Professional Documents
Culture Documents
1, January 1997
B. L. Harshe W. Z. Black
Consumers Power Company George W. Woodruff School of Mechanical Engineering
Jackson, Michigan Georgia InstitMe of Technology
Atlanta, Georgia
ABSTRACT - A mathematical thermal model is developed judgement. Analytical work at San Diego Gas & Electric in
to predict the operating twnperatures of cables in a single 1970 led to a thermal model proposed by Stolpe [1] which
covered tray when there is load diversity in the cable provided the basis for the tunpacities found in the ICEA/NE-
bundle. The model accommodates two different loading MA Standard P54+40 [2], This standard is very conserva-
scenarios - one in which the heat is distributed evenly tive because it assumes that every cable in the tray is loaded
across the cable tray cross section and the second one at the maximum code allowable current. The conservatism
which concentrates the heavily loaded cables along the in Stolpe’s model assures that every cable in” the bundle
centerline, while surrounding them with more lightly operates at a temperature that is less than a selected maxi-
loaded cables. The temperature predictions provided by mum vahte. This work provided the, first basis for establish-
the model are compared to data found in other IEEE ing cable ampacities based on experimental results, however,
papers, data collected in laboratory measurements, and it was limited to open top cable trays. Since the introduction
new data from a four-year study of cable trays in an of P54-440, several stales that are primarily experimental in
operating nuclear plant. Reasons for differences between nature have been used to establish the ampacity of cables in
the field data and the computer results are discussed. The covered trays.
model is used to evaluate the conservatism in the available The best guidance for selecting a derating factor for cables
Codes and Standards. in covered trays comes from two sources. The first is the
A derating factor is introduced that is defined in terms National Electric Code [3], and the second is an IEEE paper
of the ampacity of cables in open-top trays. The derating by Engmann [4]. The National Electric Codo introduces a
factor accounts for the added thermal resistance present derating factor baaed on cable size and severely limits the
when a cover is placed over the cables, trapping a layer of amount of cable that can be placed in a tray. Engmatm [4]
stagnant air on top of the cable mass. The computer suggests a constant derating factor for all size cables and all
model is then used to predict values for the derating fill conditions covered by ICEA/NEMA Standard P54440[2].
factor as a function of cable depth. The detiating factor The problem of selecting an appropriate derating factor for
is shown to be independent of the composition of cables in cables in covered trays is complicated when diversity exists
the tray. The presence of a cover is shown to reduce the in the cable loading. Studies [5] on the effect of dhrerse
ampacity based on an uncovered tray by up to 25 percent loading on the cables in open top trays have shown the large
depending on the depth of the cables in the tray. amount of conservatism in the Code for practical conditions.
Rarely are cable trays loaded in the manner amumed by the
I. INTRODUCTION Code, because the cables in a tray do not simultaneously
carry the maximum code allowable current for an extended
Before 1974, the derating of cables in open-top cable trays period. Typically, the actual electrical load is significantly
was calculated on a case-by-case basis using engineering less than the maximum code allowable value due to conserva-
tive cable sizing practices used by design engineers. There-
fore, the actual temperatures of cables in trays are typically
lower than those predicted by ICEA/NEMA Standard P54-
440. This fact suggc@s that the cables maybe conservatively
96 W 209-7 PWRD A paperrecommendedand approvedby the IEEE operated at current levels above those presently allowed by
Insulated Conductors Committee of the IEEE Power Engineering Society for the Code.
presentation at the 1996 IEEWPES Winter Meeting, January 21-25, 1996,
Currently; there is no consistent guidance that can be
Baltimore, MD. Manuscript submitted March 27, 1995; made available for
printing November 16, 1995. reliably used to address the derating of power cables in
covered cable trays, especially when diversity is present in
the loading of the cables. This paper provides a model that
can be used to calculate derating factors for many cable tray
geometries. The model results are compared with test
measurements provided in other studies and by an extensive
or, (7)
when,
Tc ew Rat <3284
h top
A I T~ e~ 1
t <hgw and h for the bottom surface of the cable bundle is [9,
pg.433]
-r ~
J4[$(TS-T
-’P
H - .,-------------- /-’””
. - . . . . . . . . . . . . . . . . . . . . . ---- (8)
h~ = 0.295—
when,
‘@’=
OolMk[$T.-T-)r (lo) After the surface temperature of the cable bundle is deter-
mined and once the distribution of heat generated within
various cables that make up the cable mass is specified,
centerline or maximum bundle temperature can be calculated.
the
the
107< Raw < 10]1 The present codes and standards are based on (the assumption
that all cables in the tray are energized to such a level that
or, the energy is distributed uniformly throughout the bundle
hm = 0.589—
cross-section. Referring, to this distribution w the uniform
J4[*C-T-)F “1)
model, the relationship between the maximuml cable temper-
ature and the surface temperature is given by [1].
&W = 0.0888w3
() + (Tc -Tm)
(12) code allowable limits. However, a simple modification
uniform assumption allows a calculation
to the
of a maximum
temperature that accounts for a diversity of cument loading in
and the Rayleigh number used in (6) and (7) is: the cables. This model, which accounta for load diversity, is
called the layered model. The layered model assumes that
Rut = 0.71t3
()
~ (T$ -Tc)
(13) the most heavily
centerline
loaded cables are placed along the bundle
where they will achieve the i highest
and they are surrounded by the more lightly loaded cables.
temperature
The thermal properties to be used in these expressions Using this assumption, the layered model provides the most
depend on temperature and they can be expressed as [8, conservative cable temperature profile, because the lightly
pg.520] loaded cables act as a blanket that insulates the more heavily
loaded cables along the tray centerline. The layered model,
k = 7.0x10-sT~ + 0.0237 (14) therefore, accounts for load diversity but retains a conserva-
tive approach to ampacity calculations by predicting a
maximum cable temperature profile in the bundle. For the
@ = 1.02x104 ~ - 2.394x106 ~ + 1.t35x10s (15) layered assumption, the relationship between the surface
V2 temperature and the maximum bundle tempemture is [5]
moderately loaded, or lightly loaded. The heavily loaded Table 1 Range of Test Condhions
cables number NH and are placed along the bundle centeq
moderately loaded cablea number N. and are placed over the
heavily loaded layer; and finally, the lightly loaded cables
number N~ and are placed on the outside the cable bundle.
For a loosely packed cable array, the depths of the three Tray Tray Depth % of % of
layers that appear in (17) can be determined by the expres- Depth Width of Fill Power ICEA
sions: inches inches inches Cables Allowable
(cm) (cm) (cm) in Ampacity*
Tray
loaded. The thermocouples placed dkectly on the heavily measurements carried out at Georgia Tech [1 O]. Three
loaded cables in the cable tray were used to measure the ladder trays were loaded with three dlffererk cable sizes. The
ms.ximum temperature within the cable bundle and to veri~ temperatures from forty thermocouples placed throughout the
the layered temperature prediction of the computer model. cable mass were recorded by an automatic data acquisition
The remainin g thermocouples were scattered throughout the system. Only the data from the bottom tray in the three cable
bundle area and in the air space between the cover and the tray stacked configuration were used. The data resulting
bundle. The thermocouple readings were averaged along from this tray closely simulated a single tray surrounded by
with the layered temperature(s) to obtain a weighted average stagnant ambient air and not heated by the cable trays below
temperature of the cable bundle. The percentage of layered it.
temperatures used to calculate the uniform temperature was
proportional to the percentage of the energized power cable Iv. DISCUSSION OF tiSULTS
area in the cable tray. This uniform temperature was used to General
validate the temperature prediction of the uniform computer
model. To determine the accuracy and conservatism of the
Thermocouples were installed longitudinally on both sides proposed thermal model in predcting temperatures in
of the test location to verify that a negligible axial tempera- realistically operated cable trays, cable temperatures were
ture gradient existed along the cable tray. The ambient air measured in an operating power plant and temperature data
temperature was measured at one inch above and below the were collected from two other studies [4 and lo] as described
cable tray at each test location. The average of these in the previous section . Each cable tray section tested was
measurements was used as the ambient air temperature. then simulated using the model. The predcted temperature
Since the heat generated within the cable bundle was low, the rises are compared with the measured temperature rises and
presence of the tray did not influence the air temperature are shown in Fig. 2. A conservative temperature prediction
measured by the thermocouple above the tray. The differ- falls below the 450 line. Nonconservative results appear
ence between the air temperature above and below the tray above the 450 line. ,’
was always less than 1“C. In Fig. 2, the data points fall below the450 line suggesting
The current was measured in all of the instrumented that the model provides conservative results,, In fact the
heavily loaded power cables in the cable tray. The current of model predicts a maximum temperature that if; often signifi-
the remainin g power cables in the instrumented trays was cantly greater than the measured value. In a few caaes, the
obtained from other sources, such as name plate data. model pre&cts the temperature rise in the cable bundle that
Control cables were assumed to have no current. Sensitivity is approximately three times the value measured by the
studies carried out during the experimental investigation thermocouples. The experimental data by Engmann [4] was
supported this assumption. collected for trays in which cables were evenly loaded to
To simulate a long run of covered tray, tray covers ten feet 90”C ampacities in a laboratory setting. These data points
in length were used. To prevent the flow of cool air into and naturally cluster near the 450 line. This behavior is expected
out of these trays, insulation was inserted into the ends of because the neatly laid cables are heavily loaded and they
each tray test section to block the flow of air. Where cables
entered or exited the tray within the tray test section, a hole
was provided in the tray cover. The hole was blocked with
insulation to prevent air movement.
provide uniform heat generation across the cable bundle cross energized cables decrease. From this observation it can be
section. Therefore, the conditions during the test more nearly concluded that the load diversity in a realistically loaded
match the assumptions used in the mathematical model. covered tray will produce significantly lower temperatures
than predicted by the existing code [2]. Fig. 3 is similar to
Effect of Load Diversity on Ampaci~ the earlier published curves [5] for uncovered cable tray, but
it includes the added thermal resistance imposed by the
The layered thermal “model can be used to quantify the presence of the tray cover.
influence of diversity on the temperature rise within a
covered cable tray. Fig. 3 shows the temperature rise Antpacity Ratings Eased on a Derating Factor
predicted by the thermal model plotted as a function of the
percent of ICEA code allowable current. This curve assumes A simple way to determine the maximum allowable
a 4“ by 24” covered’ tray that is 75 percent filled (3-Inches) ampacity of cables in covered trays, is to determine the
with #10 AWG cables. When a cable tray contains several required reduction in ampacity that results from the presence
power circuits at various percentages of code allowable of the tray cover. This reduction in ampacity C= hen be
ampacities, it should be assumed that all of the cables are dkectly related to the ampacity values for cables in uncovered
loaded at the worse case level when using Fig. 3. For trays published in existing standards. Using this philosophy,
example, if power circuits account for 25 percent of the total a derating factor can be developed. Once values for the
cable energiti area, and the worse case Ioadmg in one set derating factors are established, the ampacity of cables in a
of the cables is 130 percent of code allowable, then all of the covered tray can then be simply obtained by applying the
power cables should be assumed to operated at 130 percent derating factor to the code allowable ampacity for cables in
of the ICEA code current. The temper@ure rise for a an uncovered tray. Using this approach, the ampacity values
covered tray for this situation would be conservatively published in the present codes, standards, and previous
estimated at 350 C as shown in Fig. 3. ampacity models that have been ustyi reliably for many years
Modeling many trays with varying fill configurations revealed can be retained.
that the cable size and the mix of the cables do not influence The derating factor is simply del%yxi as the ratio of the
the curves generated. Therefore, even though the data in maximum code allowable ampacity of a cable in a covered
Fig. 3 were developed for a tray loaded with a single size tray to that of the same cable in an uncovered tray. This
cable size, the results can be used for a tray that contains a definition assumes that the uncovered and covered trays are
variety of cable sizes. The indlvidurd curves in Fig. 3 plot identically loaded, have reached the same nyximum operating
the results assuming only that proportion of the cable in the temperature, and have experienced the same ambient air
covered tray is energized. All other cables in the tray are temperature. The derating factor is defined as:
assumed to be either deenergized or generate a negligible
amount of heat. The percent energized value is the percent
energized cable area in the tray. The results show that the ‘rlltcowed
temperature rise dramatically decreases as the percent of
Since the two trays are assumed to be loaded with identical
cables, this derating factor can also be expressed in terms of
90
the rate of heat generated in the covered and uncovered trays
100%ENERGIZED
so or the derating factor can be written as:
G 75%ENERGIZEO /
~ 70 .-----__---- /
e
~ —————-
,/
/’
‘F=fmEi
Equation (20) provides a simple way to calculate the
(20)
1s0
/
/ /
/
G 140 /
<.
1-
81~
i
--y”””’
covered
/
///
//
////
L_Tm
Fig. 4 Development of Derating Factors Fig. 6 Thermal Circuit for Covered Tray
10
to that of conduction. At this point, (7) dominates and What is more important, it will limit the selection of a
improved heat transfer results., For trays with large fill derating factor to one that requires only a knowledge of the
levels, the thickness of the air gap strongly influences the depth of the cables in the tray.
total resistance in the thermal circuit. The thickness of the
air gap can ultirpately improves the derating factor in these V. CONCLUSION
cases, as indicated in (7).
Since the der@g factor is most strongly affected by the The computer model proposed in this paper conservatively
depth of cables and the tray height, it is reasonable to plot the pre&cta cable temperature rise in covered trays. It accounts
derating factors as a function of these two variables. The for diversity in the loading of the cables within the covered
computer model described in the first part of this paper was trays and provides a theoretical justification for cable derating
used to construct curves like the ones shown in Fig. 4 for a factors that are not present in the National Electric Code.
given tray size and cable depth. From these curves the The model confirms that the suggested derating factors from
derating factors were calculated. Fig. 7 shows a composite previous investigators are conservative, and it may be
of the derating factors as a function of cable depth for three permissible to relax them using conservative criteria. The
different tray heights. The curves show that the derating derating factor is shown to be independent of cable size and
factor is not a function of the tray height provided the a conservative value for the derating factor that is a function
thckness of the tir gap above the cables is greater than one- only of the cable depth in the tray is recommended.
half inch (13mm). Once the cable depth approaches the tray
height, however, the derating factor increases because the VL ACKNOWLEDGEMENTS
thermal resistance of the air layer dramatically decreases (7).
This behavior explains the three, nearly vertical “tails” on the The authors thank the Consumers Power Company System
curves in Fig. 7. Protection Department and the Palisades Nuclear Power Plant
The application of the derating factor to the ampacity of Electrical Maintenance Department, the Electrical Systems
cables in covered trays can be fhrther simplified by ignoring Engineering Department, and the Electrical Design Engi-
the’ influence of the tray height on the derating factor. Since neering Department for their assistance. The authors also
choosing a lower value for the derating factor provides want to thank Carl Fox of the Palisades Configuration
conservative ampacity values, the derating factors suggested Control Project for his work in coordinating the test program;
by the “tails” of the three curves in Fig. 7 can be ignored. Daniel Ugorcak for his extensive technical analysis and
As explained above, these ‘tails” occur only when the cable review; and Jay Blewett for his review in support of. the
depth comes to within.one-half inch of the tray height. Using publication of this document.
the lower extreme of the derating factor curve, a consemative
derating curve based solely on the cable depth can be
established. The result of this is the recommended derating
factor curve’ shown in Fig. 8. Using the derat@g factors in
Pig. 8 will always produce conservative ampacity values.
o s (m ,0 16
0.9 I 1 I I 1 1 1 , 1 1 I (
4 i
a?~, o.7~
o 1 2 4 6 e 7
C~e Oepth (inches) 0able30epth (k+@
Fig. 7 Derating Factors as a Function of Cable Depth for Fig. 8 Recommended Conservative Derating Factors for
3-Inch, 5-Inch, and 7-Inch High Trays Covered Trays as a Function of Cable Depth
11
B. L. Harshe and W. Z. Black: The testing was performed on a wide range of trays with a
variety of fills consisting of mixtures of power cables of
The thermal model for conduction of heat through the cable varying sizes mixed with control cables. The measured
mass, and the convection and radiation effects that occur from temperatures did verifi that the temperature rise is a function
the exposed surfaces to the ambient air is simplified by a only of the amount of hwat generated in the cables and not a
number of assumptions stated in the paper. The bundle of function of cable size. It was found that unless the cables
cables consisting of conductors, insulating materials and were carrying over 250/. of the code allowable value, the heat
entrapped air between the cables is replaced by a uniform generated in either power cables or instrument cables did not
mass with an equivalent thermal resistivity that approximates significantly affect bundle temperature.
14
Temperatures were measured with 20 AWG thermocouples Several regions of test tray were located in the vicinity of
that were attached to the outer surface of the cables. The circulation fans. No data were collected when these fans were
conductors operated at slightly higher temperatures, but as in operation. Therefore all data shown in the paper were
explained in [5] the temperature drop across the insulation collected when free convection conditions existed.
material has been shown to be rather small. The
thermocouples could not be attached to the conductors The convection coefficient used for the bottom surface of the
because the tests were carried out on cable installed in an tray is based upon heat transfer from a downward facing
operating nuclear plant. The impact of field variables was a horizontal heated plate. The cable mass is assumed to be in
major concern to the investigators, and the way they were good thermal contact with the bottom of the tray and the
addressed was covered in detail in [5]. The testing for this thermal resistance of the thin, metallic tray is neglected.
paper was performed during the same time frame as the Therefore we do not need to include an additional conduction
testing for [5]. If there was any reason to doubt the validity of resistance term in the thermal model.
the data (air movement present, non-steady state heat transfer
present, or the ztemperature monitor was found out of
calibration, for example), the data was discarded. Manuscript received April 17, 1996.